JPH0385026A - Adaptive automatic equalizer - Google Patents

Abstract

PURPOSE:To improve the reliability by judging whether or not a transmission line distortion exceeds the equalizing capability of an adaptive automatic equalizer, and equalizing a data signal only when the distortion does not exceed, thereby preventing the equalizer from giving for itself distortion further onto the data signal. CONSTITUTION:A mean value measuring section 23 measures a mean value of an equalizing error signal e(t) for a predetermined period within a training period. Then a comparator section 24 compares a measured mean value with a predetermined threshold level. Then whether or not a transmission line distortion exceeds the equalizing capability of an equalizer is discriminated momentarily at the end of the training period. When a mean value is larger than the threshold level, that is, the transmission line distortion exceeds the equalizing capability of the equalizer, a data signal in succession to the training signal is outputted as it is from an output terminal out without being equalized by the changeover of a changeover switch 21.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Field of Industrial Application) The present invention relates to an adaptive automatic equalizer that automatically compensates for transmission line distortion.

(Conventional technique +117) Figure 3 shows the digital input to the adaptive automatic equalizer) [3
This shows an example of the configuration of signal 3 in Eq.

As can be seen from FIG. 3, the signal 3 is composed of a training signal 1 of a certain length and a data signal 2. The configuration of a conventional adaptive automatic equalizer to which such a signal 3 is input is shown in FIG.

The equalizer in FIG. 4 automatically switches between the training period and the decision feedback period in response to the input of signal 3. During the training period, the movable terminal 14A of the switch 14
is switched to the fixed terminal a side, and the internal parameters (tap coefficients) of the filter 11 are updated using the equalization error signal based on the training signal 1. In the decision feedback period, symbols are estimated and output based on the data signal 2 that has passed through the filter 11 whose tap coefficients have been updated in advance during the training period. At this time, the filter 11
The tap coefficient of is finely adjusted.

More specifically, the signal 3 is subjected to distortion in the transmission path, sampled at appropriate time intervals, and is supplied to the filter 11 as the equal flower input signal "(t). The output signal 1 of the filter 11
(t) is input to one input terminal of the differentiator 15 during the training period. A known training signal (equalizer output signal 1(t)) output from the training signal generating section 13 is input to the other input terminal of the differentiator 15 via the switch 14. The differentiator 15 generates a geometric error fd e (t) as the difference between the two signals 1 (t) and I (t), and applies it to the tap coefficient control section 16 . The control unit 16 updates the tap coefficients of the filter 11 according to the signal e (t). After this operation is repeated several times based on the training signal 1, a decision feedback period begins. In the decision feedback period, the input signal r (t) based on the data signal 2 passes through the filter 11 whose tap coefficients have been updated, and the output signal ! (t) and reaches the identification section 12. The identification unit 12 estimates the symbol based on the input signal I(1), and outputs the equal flower output 1.
Output as (t). Also in this decision feedback period, the tap coefficients of the filter 11 are finely adjusted by the output (e(t)) from the differentiator 15.

Thus, the training period precedes the decision feedback period, and the adaptive automatic equalizer
By converging the internal parameters, that is, the tap coefficients of the filter 11, the transmission path distortion experienced by the data signal during the subsequent decision feedback period is compensated.

(Problems to be Solved by the Invention) However, when transmission path distortion increases for some reason, the tap coefficients of the filter 11 may not be able to converge depending on training within a predetermined training period. In such a case, in the conventional adaptive automatic equalizer described above, stability during the decision feedback period deteriorates, and the equalizer becomes in a divergent state. As a result, there is a problem in that the equalizer adds distortion that is more serious than the transmission path distortion, increasing the possibility that the reliability of the equalizer output will be significantly degraded.

The present invention has been made in view of this point, and its purpose is to determine whether or not the transmission path distortion exceeds the equalization capability of the adaptive automatic equalizer, and to equalize the data signal only if the distortion does not exceed the equalization capability of the adaptive automatic equalizer. However, if it exceeds the level, the data signal is output as is without being equalized, preventing the equalizer itself from adding further distortion to the data signal, and improving the reliability of the equalizer output. The aim is to improve

[Structure of the Invention] (Means for Solving the Three Problems) The first adaptive automatic equalizer of the present invention inputs a signal having a training signal and a subsequent data signal, and eliminates the transmission path distortion caused by this signal. a filter for compensation; an identification unit that inputs the output from the filter and estimates and outputs the symbol of the transmitted digital signal; and an equalization unit that is the difference between the output of the filter and the output of the identification unit. a control unit that sequentially corrects internal parameters of the filter using an error signal, and after a training period that converges internal parameters of the filter using an equalization error signal based on the training signal, the data Adaptive automation in which the identification unit estimates and outputs a transmission symbol of the data signal passing through the filter during a decision feedback period in which internal parameters of the filter are successively corrected using a geometric error signal based on the signal. In the equalizer, an average value measuring means for calculating an average value of the equalization error signal within a predetermined period within the training period; and an average value of the equalization error signal and distortion compensation by the filter. a comparison means for comparing the data signal included in the equalization input signal with a predetermined threshold value for determining pass/fail when the average value of the equalization error signal is larger than the threshold value; and a switching means for switching to output without equalization.

In the second adaptive automatic equalizer of the present invention, in the first adaptive automatic equalizer, the filter includes a plurality of taps, and the controller corrects the internal parameters of the filter by tapping the respective taps. When the average value is greater than the threshold value, the switching means fixes the tap coefficient of a certain one of the taps to 1, and changes the tap coefficient of the other taps to 1. is fixed at O.

(Function) In the first and second adaptive automatic equalizers of the present invention, the average value measuring means measures the average value of the equalization error signal during a predetermined period within the training period. Next, a comparison means compares the average value with a predetermined threshold value. As a result, at the end of the training period, it is instantly determined whether the transmission path distortion exceeds the equalization ability of the equalizer. When the average value is larger than the threshold value, that is, when the transmission path distortion exceeds the equalization ability of the equalizer, the data signal subsequent to the training signal is not equalized by switching the switching means. It will be output as is.

In the second adaptive automatic equalizer of the present invention, the switching by the switching means is performed, for example, as follows. That is, in the second adaptive automatic equalizer, the tap coefficient of a certain one of the plurality of taps of the filter is set to 1.
, and the other tap coefficients are fixed to 0. By switching these, the data signal is output without being equalized. This prevents new distortion from being added to the data signal.

(Example) The present invention will now be described in detail with reference to the drawings.

FIG. 1 is a diagram showing a first embodiment according to the present invention.

In FIG. 1, circuit components equivalent to those in FIG. 4 are given the same reference numerals as in FIG. 4. The differences between FIG. 1 and FIG. 4 are as follows. That is, the second switch 21 is connected in series between the filter 11 and the identification section 12 in FIG. The second switch 21 has a movable terminal 21A
is fixed terminal a.

b can be switched to either side. This switch 21
The switching is performed by the average value measuring section 23 and the comparing section 24 connected to the output stage of the difference device 15. That is, the average value measurement unit 23 measures the absolute value of the equalization error signal e (t) output from the difference device 15, and the average value. 1. It is preferable that the measurement period for the average value is from a certain point in the middle of the training period to the end of the training period. The time at which the measurement is started is preferably the time required for the tap coefficient control unit 16 to update the tap coefficients of the filter 11 twice as many times as the number of taps included in the filter 11.

During such a measurement period, the average value of the signal e (t) obtained by the average value measurement section 23 is added to the comparison section 24 . The comparison unit 38 compares the given average value with a predetermined threshold. For example, this threshold is equal to the vase output!
(t) is determined to be useful for determining whether the bit error rate of (t) is larger or smaller than a desired value. In other words, this threshold value is determined by the distortion compensation by the filter 11,
That is, it is determined as a standard for determining whether or not distortion has been newly added by passing through the filter 11. The second switch 21 is switched depending on the comparison result. That is, when the above-mentioned average value is smaller than the threshold value, the switch 21 remains switched to the terminal side during the judgment feedback period after the training period. As a result, from the output terminal OUT, the equalizer output signal of the geometric neighbor! (t) will be output. On the other hand, when the average value is larger than the threshold value, the switch 21 is switched to the terminal a side of the bypass path 22. As a result, in the decision feedback period after the training period, the equalizer output 1 (t) that has not passed through the filter 11 is output from the output terminal OUT.

In this way, if the parameters of the adaptive automatic filter (c) section, that is, the tap coefficients of the filter 11, cannot be converged during the training period preceding the decision feedback period, the second switch 21 switches to the bypass path. 22 side, and the equalizer itself no longer adds significant distortion to data No. 12 in the decision feedback period.

FIG. 2 is a diagram showing a second embodiment according to the present invention.

In FIG. 2, the same components W as in FIG. 1 are given the same reference numerals as in FIG. The differences between FIG. 2 and FIG. 1 are as follows. In Fig. 2, a transversal type filter with five taps is used as the filter, and a so-called decision feedback type (Decision Feedb) is used as the filter.
ack Equalizer) was used. The transversal filter 11 is
Delay units 11a to 11d; multiplication units 11A to 11D that multiply the signal delayed by the delay unit 11 by a tap coefficient;
It is composed of an adder 11α that adds the outputs of these multipliers.

The apparatus shown in FIG. 2 operates in substantially the same manner as the apparatus shown in FIG. 1, but the differences are as follows. During the training period, the known training signal from the training signal generator 13 is transmitted to the delay unit 1 of the filter 11.
It is also added to 1d (llc). As a result, adder 1
1α contains the following five signals: a signal in which the equalizer input signal (training signal 1) r (t) passes through the multiplier 11A and is multiplied by a tap coefficient, and a signal in which the training signal 1 is multiplied by the tap coefficient in the delay unit 11a. After being delayed by one sample time interval j, the signal multiplied by the tap coefficient passes through the multiplier 11B, and the training signal 1 is transmitted to the delay sections 11a and l.
multiplier 11 after being delayed by two sample time intervals in lb.
The signal multiplied by the tap coefficient through c and the known training signal from the training signal light generation section 13 are delayed by one sample time interval in the delay section lid, and then passed through the multiplication section 11E and multiplied by the tap coefficient. The signal and the known training signal are delayed by two sample time intervals in the delay units lid and llc and then passed through the multiplication unit 11D to receive a signal multiplied by a tap coefficient. These 5
A sum signal (I(t)) of the two signals is added to one input terminal of the difference S15. A known training signal from a training signal generator 13 is applied to the other input end via a switch 14. As in the case of FIG. 1, the differentiator 15 converts the difference between the two inputs (.
Output as (t). Based on this signal e (t), the tap coefficient control unit 16 controls the multiplication units 11A to 11 . of the filter 11 after the training period. Update the H tap coefficient.

This update is performed every time a sample value (r(t)) is input. Furthermore, the average value measuring section 23 and the comparing section 24 operate in the same manner as in FIG. 1 based on the output (e(t)) from the differentiator 15. However, in FIG. 2, the output of the comparison section 24 is applied to the tap coefficient control section 16. The operation of the tap coefficient control section 16 is controlled by the comparison section 24.
The results of the comparison differ depending on whether the above-mentioned average value is smaller than or larger than the above-mentioned threshold value. When you're small,
The tap coefficient control unit 16 finely adjusts the tap coefficients of the multipliers 11A to 11E based on the output of the differentiator 15 during the decision feedback period. As a result, the five signals output from the multipliers 11A to 11E are added to the adder 11α. The output (1(t)) from the adder is symbol-estimated by the identification unit 12, and sent to the output terminal OUT via the switch 14.
is output from. Further, when the average value is larger than the threshold value, for example, the multiplier 11A.

1. Fix the tap coefficients of 1B, IID, and IIE to O,
The tap coefficient of the multiplier 11C is fixed to 1.

As a result, during the decision feedback period, each tap coefficient is fixed as described above, and therefore, the adder 11α receives only the output from the multiplier 11C, that is, the sample 1ii(r
(t)) is added as is.

As described above, according to the embodiments shown in FIGS. 1 and 2, it is instantly determined whether or not the transmission path distortion exceeds the equalization capability of the adaptive automatic equalizer, and the adaptive automatic equalizer itself Further, it is possible to prevent new distortion from being added, and there is no need for a memory to store the equalizer input signal r (t) for a certain period of time.

In the above embodiment, only a decision feedback type adaptive automatic equalizer was shown, but the present invention can also be applied to a linear adaptive automatic equalizer.

Note that a complex number can also be used as the isometric input signal r (t). In this case, the above-mentioned isolator error signal e (t) becomes a complex number, and the average value determined by the average value measuring section 23 based on it becomes a plurality of absolute values.

〔Effect of the invention〕

As detailed above, according to the present invention, when the transmission line distortion exceeds the equalization capacity of the adaptive automatic equalizer, there is no need for a memory to store the equalizer input signal for a certain period of time. This fact can be instantly recognized at the end of the training period, and the reliability of the output can be improved by preventing the controller itself from adding new distortion to the data signal.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing one embodiment of an adaptive automatic equalizer according to the present invention, FIG. 2 is a block diagram of a different embodiment of the present invention, and FIG. 3 shows the configuration of signals handled by the adaptive automatic equalizer. FIG. 4 is a block diagram showing the configuration of a conventional adaptive automatic equalizer. 1... Training signal, 2... Data signal, 11
...filter, transversal filter, 12...
- Identification unit, 13... Training signal generation unit, 14.
. . . Switch, 15 . . . Differentiator, 16 . - Delay section, 11A to 11E... Multiplication section, 11α... Addition section.

Claims (1)

[Claims] 1. A filter that receives a signal including a training signal and a subsequent data signal and compensates for transmission path distortion caused by this signal; and a digital signal that is transmitted by inputting the output from the filter. an identification unit that estimates and outputs the symbol of the filter, and a control unit that sequentially corrects internal parameters of the filter using an equalization error signal that is the difference between the output of the filter and the output of the identification unit, After a training period in which internal parameters of the filter are converged using an equalization error signal based on the training signal, a determination is made to sequentially correct internal parameters of the filter using an equalization error signal based on the data signal. In the adaptive automatic equalizer, the identification unit estimates and outputs a transmission symbol of the data signal passing through the filter in a feedback period, and the equalization is performed within a predetermined period within the training period. an average value measuring means for determining the average value of the error signal; a comparing means for comparing the average value of the equalized error signal with a predetermined threshold value for determining the quality of distortion compensation by the filter; and a switching means for switching to output the data signal included in the equalized input signal without equalizing it when the average value of the equalized error signal is larger than the threshold value. Adaptive automatic equalizer. 2. The filter includes a plurality of taps, and the controller corrects the internal parameters of the filter by changing a tap coefficient of each tap, and the switching means is configured to adjust the average value to the threshold value. Adaptive automatic equalization according to claim 1, characterized in that when the adaptive automatic equalization is larger than that, the tap coefficient of a certain one of the taps is fixed to 1, and the tap coefficients of other taps are fixed to 0. vessel.